|dc.description.abstract||The potential for mine wastes to generate elevated concentrations of solutes including metals, sulfate, and reduced pH exists wherever mine-waste rock is stockpiled at the Earth’s surface representing one of the world’s largest environmental problems. The assessment of the long-term geochemical evolution of mine wastes is of critical importance in the process of mine-life planning because of the potential for adverse impacts of released solutes and low pH effluent to receiving environments. The Diavik Waste Rock Project included laboratory and field experiments investigating the geochemical evolution of low-sulfide mine-waste rock at different scales. The experiments included small-scale humidity cells (0.1 m high; laboratory), medium-scale lysimeters (2 m high; field), and large-scale test piles (15 m high; field) to facilitate development of a mechanistic approach to scaling results of the laboratory experiments to make assessments regarding the geochemical evolution at the larger field experiments. This process, generally referred to scale-up, often involves the use of humidity cell experiment results coupled with empirical scale factors to make predictions about the long term geochemistry of effluent released form mine-waste stockpiles. The empirical factors used typically include parameters known to influence rates of sulfide oxidation including mineral content, particle-size distribution, temperature, moisture content, and oxygen availability. These scale-up factors often fail to account for site specific heterogeneities in physical and chemical properties that can strongly influence the prediction process. Mechanistic approaches (i.e., the use of geochemical models including reactive transport models) have the potential to include complex heterogeneities that facilitate a quantitative assessment of the long-term geochemical evolution of mine wastes.
A conceptual model of the geochemical evolution of low-sulfide waste rock was developed to facilitate numerical simulations of the small-scale experiments and then was used to simulate the geochemical evolution in the larger scale field experiments. The conceptual model, based on oxidation of sulfide minerals coupled with the geochemical weathering of host minerals present in waste rock produced at the Diavik Diamond Mine (NT, Canada), was implemented using the reactive transport code MIN3P. The 1-D model was calibrated to capture the effluent concentrations from the laboratory-scale experiments then used to simulate the geochemical evolution at the larger scale field experiments, without further calibration, to assess the efficacy of the mechanistic scale-up approach. Geostatistical analyses of mineralogical and particle-size distribution samples were conducted to assess the heterogeneity of S, C, and saturated hydraulic conductivity. The results of the geostatistical analyses were used to inform spatial distributions of S, C, and saturated hydraulic conductivity as input to reactive transport simulations of the large-scale field experiment. The 2-D simulations were conducted to assess the influence of heterogeneity in S, C, and saturated hydraulic conductivity on the geochemical evolution of the waste rock.
The results of the humidity cell simulations indicate that the conceptual model represents the primary geochemical processes of the low-sulfide waste rock weathering. The simulated effluent concentrations compares well with the measured solute concentrations from the humidity cells, although some divergence for specific parameters was observed. Mineral surface area, mineral content, temperature, and pH were identified as important factors controlling the geochemical evolution of the waste rock. The results of the model developed and calibrated at the humidity-cell scale suggested that the conceptual model could be representative of the DWRP waste rock weathering in general and the implemented model could be used to simulate waste rock weathering for the field scale experiments.
The implementation of the conceptual model at the medium-scale field experiments involved inclusion of measured temporally dynamic temperature and infiltration to better represent the physical conditions at the field experiments. Implementation at the large-scale test pile experiments included providing spatially dynamic temperature. Inclusion of these parameters as model input facilitated completion of multi-year simulations essential to making long-term assessments of the geochemical evolution of waste rock. Scaling the humidity cell conceptual model to simulate the geochemical evolution at the field-experiment scales resulted in good visual agreement between measured and simulated concentrations and mass flux of most parameters. The pH was generally over estimated in the medium- and large-scale field simulations. Supplemental simulations indicate that calcite availability was lower for the field experiments (approximately 20% of measured content).
The field experiment simulations did not rely on geochemical data for calibration; however, these simulations did rely on site-specific physical data, including mineralogy-related parameters such as volume fraction, hydrology-related parameters including hydraulic conductivity, grain-size distribution, porosity, and water-retention curve values, and environmental parameters including temperature, precipitation, and O2 concentration (the field systems were not O2 limited); to facilitate an assessment of the geochemical evolution of waste rock. The reactive transport simulations demonstrated that a comprehensive, integrated conceptual model representing the geochemical evolution of low-sulfide waste rock, implemented and calibrated at the humidity-cell scale can be applied to field-scale experiments using a small number of measurable parameters to constrain the simulations. Parameters should include mineral content, bulk mineral surface area, and particle-size distribution, water flow and infiltration characteristics as well as general climatic conditions (specifically temperature and precipitation). The reliance on readily available, measurable parameters suggests that this approach could be implemented at other sites using the appropriate site specific parameters. This mechanistic approach provides the basis for predictive scale-up.
Consideration of the influence of temperature on the geochemical reactions was a major factor facilitating the scale-up of the model. The humidity cell experiments were conducted at temperatures 5 °C and 22 °C to allow calibration for the influence of temperature, which was a critical component in the scale-up process because of the varied temperatures at which surface-stored waste rock is exposed. Measurement of temperature at the field scale would be an important component of any scale-up program.
The results of the geostatistical analyses indicate the spatial distributions of S, C, and saturated hydraulic conductivity in the test-pile experiments could be approximated using a log normal distribution with mean and standard deviation calculated from samples collected during test pile construction for each parameter. A lack of spatial dependence for matrix hydraulic conductivity was significant because the matrix material exerted strong control over the flow of water through the test-pile experiments. The spatial distributions of S, C, and saturated hydraulic conductivity in the test piles experiments provides a foundation from which full-scale waste-rock piles could be characterized using the geostatistical methods described. The spatial dependence of saturated hydraulic conductivity in larger piles may also depend on the influence of features which were not present in the test-pile experiments (e.g., traffic surfaces).
The investigation of the influence of mineralogical and physical heterogeneity on the geochemical evolution in the Type III test-pile suggested that heterogeneous distributions of S and C mineralogy and saturated hydraulic conductivity field resulted in variations of effluent concentrations that were at times, consistent with the measured variation. Analysis of the results of the heterogeneous simulations indicate that the distribution of solute mass fluxes from the test-pile experiment for most parameters could be best approximated with a log normal probability density function.||en